The HotelWiFiTest website has added a seemingly clever new feature that claims to predict the wireless Internet (WiFi) speed you will receive in most major hotels around the world, which can be used alongside data collected from existing speedtests to help people pick a hotel that suits their needs. But there may be trouble ahead..
The site, which first launched in October 2013, currently lists hotels alongside a predicted WiFi speed score collected using end-user speedtests and this is usually reflected as a range based on those tests. A “confidence” score is also used to help reflect how reliable that score is based upon the number of tests conducted and how recently they occurred etc.
Advertisement
By comparison the new estimation facility claims to use a “state-of-the-art machine learning algorithm” to predict performance, even for hotels where no speedtests have yet taken place. Apparently this works by allowing for the “determination of accurate weights, as well as the subtle, unexpected relationships between different hotel features that can affect hotel WiFi performance“. Don’t worry, we also have no idea what that means.
Unfortunately, despite using a lot of snazzy language, HotelWiFiTest isn’t able to say clearly how their “informed estimations” actually work, although they do pledge to replace any estimates with data gleamed from actual tests in the future. We ran a few tests of our own on hotels where we had struggled to get a usable WiFi, both in public areas and in-room, and in all cases the estimate predicted considerably higher speeds than were ever possible for us.
Furthermore there are some caveats which the site could do with reflecting, such as whether or not the wifi network is only available in public areas or rooms (without this people could easily be confused). Separately we note that the website chooses not to reveal any details about its owners (usually a bad sign – even their domain registration details are protected by WhoisGuard), although all of their affiliate links seem to go through the Hotels.com database.
Overall we love the concept but more information is required. But if this is ever cleared up and improved then it could become a very useful tool indeed.
Advertisement
Comments are closed